Soil Moisture Measurement in Heterogeneous Terrain

Reliable soil moisture measurement over large areas is much needed for both hydrologic modelling and remote sensing applications. For collecting such data, portable electronic sensors offer a practical alternative to gravimetric measurements. The conversion of the measured electrical output to soil moisture is nonetheless a non-trivial task as it depends on soil type and temperature. In this study, different calibration approaches of the Stevens Hydraprobe (R) soil dielectric sensor operating at 5 0 MHz are tested with National Airborne Field Experiment (NAFE) data. The objective is to evaluate the impact of soil type and temperature on the sensor response and test the applicability of a general calibration equation. During NAFE, a spatially enabled platform (Hydraprobe (R) Data Acquisition System, HDAS) was used to collect extensive measurements of near-surface soil moisture. HDAS is a handheld system integrating the soil dielectric sensor and a pocket PC with GPS receiver allowing for direct storage of location and measurement with GIS software. HDAS measurements are composed of the dielectric constant (DC) of the soil-water mixture, soil temperature, soil moisture content, salinity and conductivity. A direct comparison between the factory calibration and gravimetric soil moisture measurements indicates that the sensor response has a loss of sensitivity at soil moisture over 25% v/v in clay. On the other hand, the real component of the measured relative DC is found to be more strongly correlated to gravimetric observations than the predicted soil moisture. Following these observations, two calibration approaches based on the measured DC were tested. The first is derived by averaging the slope obtained with various soil types (general equation). The second uses the ratio of the imaginary to real component of DC (loss tangent) to describe the difference in soil properties (loss-corrected equation). Results indicate that the calculated loss tangent is able to explain most of the variability among soil types. The root mean square error (RMSE) of the predicted soil moisture is decreased from 4.0% v/v with the general equation to 3.3% v/v with the loss-corrected equation. A third-order polynomial regression between the factory equation and observations gives the best overall accuracy with a RMSE of 2.7% v/v. However, the loss-corrected equation is more robust as it does not loose sensitivity above 25% v/v and is more reliable than the soil type dependent polynomial regression. Previous analyses have shown that the Hydraprobe sensor is also sensitive to soil temperature. In this study, the temperature effect on the real component of the measured DC is evaluated with sand and clay in different moisture conditions. With sand, temperature is found to have a negligible effect with the largest effect on real DC for a 15 degrees C temperature increase (relative to 25 degrees C) of about -0.6, corresponding to a soil moisture change of about -1% v/v. With clay, the observed temperature effect on DC for a 15 degrees C increase is about 2 at 30% v/v and 4 near saturation, corresponding to a soil moisture increase of about 3% v/v and 5% v/v respectively. It is also found that the factory temperature correction algorithm increases the temperature effect on the measured real DC. Consequently, a new correction is derived based on the loss tangent, to account for different temperature effects according to soil type. The loss-corrected equation, including the proposed correction for temperature, is finally applied to the data from the National Airborne Field Experiment. Maps of soil moisture at 25 0 m resolution over an area of 27 km(2) are presented for three sampling days following a rainfall event. Such spatial data will be available for calibration/validation of hydrologic models, remote sensing of soil moisture and understanding controls on spatial patterns in soil moisture.